Small hydroxyethylene-based peptidomimetics inhibiting both HIV-1 and C. albicans aspartic proteases

Article abstract of DOI:10.1016/j.bmc.2003.08.004

We have extended a highly flexible method for rapidly assembling aspartic protease inhibitors to produce symmetric and asymmetric monohydroxyethylene peptidomimetics. This method is based on the prior synthesis of the central non-cleavable peptide-bond is

Full text of DOI:10.1016/j.bmc.2003.08.004

Bioorganic & Medicinal Chemistry 11 (2003) 4719–4727
Small Hydroxyethylene-Based Peptidomimetics Inhibiting
Both HIV-1 and C. albicans Aspartic Proteases
Alessandro Tossi,a,b* Fabio Benedetti,b,c Stefano Norbedo,b,c Damiano Skrbec,a
Federico Bertib,c and Domenico Romeoa,b
aDepartment of Biochemistry, Biophysics and Macromolecular Chemistry, University of Trieste,
Via Giorgieri 1, I-34127 Trieste, Italy
bCentre of Excellence on Biocrystallography, University of Trieste, Via Giorgieri 1, I-34127 Trieste, Italy
cDepartment of Chemical Sciences, University of Trieste, Via Giorgieri 1, I-34127 Trieste, Italy
Received 20 June 2003;accepted 5 August 2003
Abstract—We have extended a highly flexible method for rapidly assembling aspartic protease inhibitors to produce symmetric and
asymmetric monohydroxyethylene peptidomimetics. This method is based on the prior synthesis of the central non-cleavable pep-
tide-bond isostere [NH2–P1cP10–NH2; c=hydroxyethylene isostere, HNCH(Bz)CHOHCH2CH(Bz)NH], with the possibility of
accurately controlling its stereochemistry (S,S,S or S,R,S), and subsequently adding appropriate flanking units, chosen from com-
mercially available amino acids, aromatic carboxylic acids, or phenoxyacetic acid (Poa) derivatives. The method was used to make
asymmetric inhibitors of general formula Kyn-Xaa-PhecPhe-dmPoa, (Kyn=kynurenic acid, Xaa=Val, Thr or d-thienylglycine,
Mr=716–754) and symmetric inhibitors of formula xPoa-PhecPhe-xPoa (xPoa=Poa or dimethyl-, hydroxy-, formyl- or acetyl-
Poa, Mr=553–609), with logPo/w values ranging from 4.1 to 7.6. Inhibition of HIV-PR did not depend on the stereochemistry of
the hydroxyl group, while it depended markedly on the substituents present on the Poa residues, with dmPoa being preferred over
Poa or its more hydrophilic derivatives. Conversely, inhibition of Candida albicans Sap2 was higher for the S,S,S epimers, and Poa
or its hydrophilic derivatives were preferred over dmPoa.
# 2003 Elsevier Ltd. All rights reserved.
Introduction
Finally, to be exploited as potential drug leads, the
protease inhibitors must possess a relatively low mole-
cular weight and a hydrophilicity/hydrophobicity ratio
suitable for a sufficient level of bioavailability.4,5,7 In this
context, we have been developing a flexible approach for
the design and synthesis of such inhibitors.14ꢀ22
Aspartic proteases play significant roles in determining
the pathogenicity of both viruses and fungi.1ꢀ12 The
design and synthesis of high affinity inhibitors of these
enzymes can thus provide useful leads for the develop-
ment of potential anti-infective drugs.1ꢀ7,11,13 A large
number of inhibitors have been characterised, but only
a handful of those inhibiting the aspartic protease of
HIV (HIV-PR), has reached clinical use.4,6,7 For the
aspartic proteases so far investigated, a high affinity of
designed inhibitors has been achieved by maximizing
the number of noncovalent contacts that they make
with the enzyme binding site. To this effect a transition-
state isostere, that is a functional group that can mimic
the tetrahedral transition-state for amide bond hydro-
lysis, is incorporated into the designed inhibitors and
appropriate flanking residues are attached to it.1ꢀ3,5,7,8,13
For HIV-PR, X-ray crystallography has shown that the
two identical chains of this dimeric enzyme are related
by a 2-fold rotation axis.1,4,7 This has led to the design
and synthesis of C2-symmetric and nearly-symmetric
compounds, 2,3,5,7,13,20 which in some cases have proved
to be very efficient inhibitors and have been developed
into drugs,4,6 presently employed in the Highly Active
Anti-Retroviral Therapy (HAART) of AIDS.23ꢀ25
Much effort has been expended in reducing the size and
complexity of these inhibitors while maintaining a high
affinity, also towards mutant proteases which have devel-
oped resistance to the first generation of drugs.7,26,27
A variety of aspartic proteases are also produced and
secreted by Candida albicans, a fungus responsible for
*Corresponding author. Tel.: +39-04-0558-3673;fax: +39-04-0558-
3691;e-mail: tossi@bbcm.units.it
0968-0896/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2003.08.004

4720
A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
Scheme 1. Overview of dipeptide isosteres available from the common precursor A.
mucocutaneous and systemic infections, frequently
encountered in immunodepressed patients such as those
infected by HIV-1.8,28 Although there are several mem-
bers of this family, probably individually expressed at
various stages of the infection process and playing dif-
ferent roles, Sap2 appears to be the major secretory
aspartic protease in log-phase Candida cells.29,30 Despite
the difference in structure between the retroviral and the
fungal aspartic proteases (Sap), which are monomeric
proteins with two equivalent domains, several inhibitors
of the HIV-1 protease are also active on Saps.8,11,21,31ꢀ33
This may explain the beneficial effect on Candida infec-
tion observed in some AIDS patients undergoing
HAART.33
assembly of inhibitors by adding appropriate flanking
residues, using standard peptide synthesis methods. The
flexibility of the method derives from the possibility of
(i) obtaining from a common precursor A (Scheme 1)
central modules with different types of non-hydro-
lysable amide-bond replacements (e.g., mono-
hydroxyethylene D, E or dihydroxyethylene B, C);(ii)
independe0ntly varying the type of side chain in positions
P1 and P1 in these central modules;(iii) precisely con-
trolling the stereochemistry of the central module;(iv)
preparing modules with either two amino terminals (1,4-
diamines B, D) or an amino and a carboxy terminal (d-
amino acids C, E)19 and (v) selectively protecting one
amino terminal, so as to permit assembly of either
asymmetric or symmetric inhibitors on addition of
appropriate flanking residues.20,22 Another elegant
method for the rapid synthesis of C2-symmetric diamino-
dihydroxyethylene modules has been reported,34 but it
does not achieve the level of flexibility of our method.
Furthermore, a key feature of our approach is the use of
off-the-shelf components at all stages, which increases
the rapidity with which novel inhibitor structures can be
assembled.
Here, we report the results of the evaluation of the anti-
aspartic protease activity of a series of diastereoisomeric
peptidomimetics, containing a hydroxyethylene isostere
as the central non-hydrolysable group, in either the
S,S,S or S,R,S configuration, and rationally selected
amino acids or organic acids as flanking units. These
compounds were prepared by a simplified, modular
synthetic approach that has been perfected in our
laboratories, with the aim to maximally reduce both the
size and complexity of inhibitors, while concomitantly
achieving a potent inhibitory capacity. The aspartic
proteases considered were a recombinant form of the
HIV-1 enzyme and native Sap2, purified from a virulent
strain of C. albicans.
We have previously integrated this synthetic methodol-
ogy with a computational method that derives the
structures of new potential HIV-PR inhibitors from a
reference compound, whose inhibition constant and
crystal structure in complex with HIV-PR are
known.14,20 Using molecular mechanics to compare
complexation energies of these new structures relative
to that of the original reference inhibitor15 it is then
possible to select the most promising ones for rapid
assembly.
Results and Discussion
Chemistry
We have developed a rapid, efficient and very flexible
Recently, we have described several small diaminodihy-
droxyethylene type inhibitors obtained in this manner,
which have IC50 values towards HIV-1 PR in the low
nanomolar range.18,20,21 These compounds confirmed
the suitability of the phenylalanine side chain in both
method
for
assembling
aspartic
protease
inhibitors.16,18ꢀ22 This method is based on the prior
synthesis of the desired type of central modules (a
pseudo-dipeptide which contains the non-hydrolysable
peptide bond isostere) (see Scheme 1), and subsequent
0
positions P1 and P1 , whereas other tested possibilities

A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
4721
stereoselectively syn-epoxidized giving the epoxyalcohol
1b in 23% overall yield from the starting aminoester.16
Reductive ring opening of 1b with Red-Al gave the diol
2 in 65% yield.22 Selective protection of the first
hydroxy group was achieved by cyclization of the N-
Boc aminoalcohol to oxazolidinone under acyl transfer
conditions (NaH/THF), while the second hydroxy
group was mesylated and the mesylate was displaced by
azide giving 3. Finally, Boc was reinstalled on the amino
group, the oxazolidinone was opened hydrolytically and
the azide was hydrogenated to give the diaminoalcohol
4 in 51% yield from 2. Alternatively, the pivotal inter-
mediate 2 was mesylated giving a bis-mesylate that
spontaneously cyclizes by a SN2 mechanism;the second
mesylate is then replaced by azide giving the oxazolidi-
none 5 in which the configuration at the ring *C(–O) is
inverted with respect to 3. The same sequence of reac-
tions already described for 3 (step f, Scheme 2) finally
gave the diaminoalcohol 6 in 32% yield from 2. Thus,
by choosing the appropriate conditions for the cyclisa-
tion of 2, it was possible to control the configuration of
the –OH group in the required isostere 4/6.22 While the
synthetic effort required to obtain 4 or 6 is comparable,
the latter was initially chosen for this study because this
is the same isostere present in ritonavir.5
Scheme 2. Synthesis of the diamminoalcohols 4 and 6, used as central
modules in peptidomimetic aspartic protease inhibitors: (a) (i)
CH3PO(OCH3)2, n-BuLi, THF, 95%;(ii) BnCHO, K 2CO3, EtOH,
50%;(b) (i) NaBH 4, MeOH, 0 ꢁC, 75%;(ii) mCPBA, DCM, 70%;(c)
RedAl, THF, 65%;(d) (i) NaH, THF, 80%;(ii) MsCl, Et 3N, DCM;
(iii) NaN3, DMSO, 18-crown-6, 80%;(e) (i) 2 equiv MsCl, DIPEA,
We have assembled a symmetric hexapeptide structure
(TS-91) by adding the Kyn-Val dipeptide to both sides
of module 6, so as to ensure complete coverage of the
enzyme binding site (Scheme 3). The pentapeptide inhi-
bitors TS-92–TS-95 are instead based on results pre-
viously obtained with dihydroxyethylene inhibitors.18,20
They have the general sequence Kyn-Xaa-4/6-dmPo0a
(Scheme 3), where Xaa is either Dtg, Thr or Val. In P2 ,
dimethylphenoxyacetic acid (dmPoa) was preferred to
Poa, in analogy with the structure of the highly suc-
cessful, new generation HIV-PR inhibitor lopinavir;26,27
this choice was also supported by the results of a com-
putational analysis.35 In fact, compound TS-94, with
the sequence Kyn-Val-6-dmPoa, differs from lopinavir
only in position P3 where Kyn replaces the cyclic urea
residue. The synthesis of TS-94 is considerably simpler
than that of lopinavir as the cyclic urea of the latter can
not be introduced as such, but requires a specific, less
direct approach.27
1,2-dichloroethane, 80 ꢁC;(ii) NaN , DMSO, 18-crown-6, 50%;(f) (i)
3
Boc2O, NaH, THF;(ii) Cs 2CO3, MeOH/H2O;(iii) H 2, Pd/C, 80%.
(e.g., Ile and cyclohexylalanine) considerably reduced
activity. This is in line with the substrate preference of
the protease2,4 and is also preferred in the clinically used
inhibitors. We also found that small natural or non-
proteinogenic amino acids, such as valine, threonine or
d-thienylglycine (Dtg) were preferred in position P2, and
carboxylic acids based on the chromone core, such as
kynurenic acid (Kyn), were suitable in position P3. Such
0
0
residues could also be placed in positions P2 and P3 ,
respectively, so as to produce symmetric pseudo-hexa-
peptide inhibitors, or alternatively the small phenoxy-
0
acetic acid moiety (Poa) could be placed in P2 to
produce smaller and asymmetric, but highly efficient
pentapeptide inhibitors.
Compounds TS-96–TS-105 are symmetric tetramers
with sequence xPoa-4/6-xPoa (Fig. 1), where xPoa
represents the series Poa, dmPoa, 4-hydroxy-
phenoxyacetic acid (hPoa), 4-formylphenoxyacetic acid
(fPoa), 4-acetylphenoxyacetic acid (aPoa). They are
small and very facile to assemble. Our intention was to
use this minimalist approach to (i) analyse the role of
central module conformation when interaction of
flanking residue with the enzyme binding-site are neces-
sarily limited, and (ii) investigate the role of substituents
on the 0Poa core in modulating interactions with the S2
and S2 enzyme binding subsites. Computational stud-
ies35 had in fact indicated that the presence of H-bond
donor/acceptor substituents on Poa could increase
complexation with the HIV-PR binding site sig-
nificantly, while also improving the hydrophobicity/
solubility characteristics.
We have now utilised this expertise for the assembly of
inhibitors with the diamminomonohydroxyethylene
type central module (Scheme 2). The synthesis of the
monohydroxyethylene central cores 4 and 622 exploits
the same key intermediates we already used for dihy-
droxyethylene pseudodipeptides,18,20 thus widening the
scope of the strategy, while maintaining all the advan-
tages described above. Briefly (Scheme 2), the methyl
ester of N-Boc phenylalanine was converted into the
corresponding phosphonoketone by reaction with lithi-
ated methyl dimethylphosphonate;Horner–Emmons
olefination with phenylacetaldehyde gave then the a,b-
unsaturated ketone 1a in 43% yield. The latter was then
reduced with sodium borohydride to give a 3:1 mixture
of diastereoisomeric alcohols;the S,R isomer was puri-
fied by column chromatography and crystallization and

4722
A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
Scheme 3. Structures of peptidomimetic aspartic protease inhibitors reported in Table 1. The stereochemistry of the non-hydrolysable isostere
hydroxyl group (R) is indicated.
0
0
The overall yield of synthesised peptidomimetics was
relatively good, ranging from 10% to over 50% from
modules 4 or 6. The principal factors limiting yields are
loss on extraction, which are significant at the limited
synthetic scale (0.1 mmol) reported here, and formation
of a side-product with flanking residues bound also to
the core hydroxy group. For symmetric inhibitors syn-
thesised in one-pot reactions, yields could be markedly
increased by directly and inexpensively purifying the
product from the reaction mixture with disposable RP-
HPLC cartridges, thus avoiding extraction. Further-
more, the side-product (estimated at 10–40% of crude)
can be efficiently reconverted to the desired product by
treatment with a base, such as 80% aqueous hydrazine.
For example, the yield of TS-96 increases from 34 to
44% after recovery. An alternative route, that of tem-
porarily protecting the hydroxy with the trimethylsilyl
group, was abandoned as it led to reduced yields due to
precipitation under our reaction conditions.
dipeptide on the P2 /P3 side of the central module with
dmPoa (TS-92–TS-95) results in a slightly increased
inhibitory activity, while molecular weight is reduced.
Varying the amino acid in P2 has little effect on IC50 (see
TS-92–TS-94), but significantly modulates the solubi-
lity/hydrophobicity characteristics as indicated by the
calculated logPo/w values (Table 1). Furthermore,
the conformation of the core hydroxy does not
seem to be critical for inhibition as indicated by the
identical activities of the epimeric compounds TS-94
and TS-95.
The negligible effect of the hydroxy group conformation
for HIV-PR inhibition is confirmed in the tetrameric
co0mpounds TS-96 and TS-97, with dmPoa in P2 and
P2 . These maintain a quite reasonable activity (ꢂ25
nM) despite a considerably reduced size, confirming the
0
effectiveness of dmPoa in interacting with the S2 and S2
binding subsites in the enzyme. That this interaction
depends on the presence of the methyl groups in this
residue is substantiate0d by the pair TS-98 and TS-99,
with Poa in P2 and P2 , for which the IC50 is shifted to
the micromolar range. In this case, in the absence of
strong int0eractions from the flanking residues with the
S2 and S2 subsites of the enzyme, the compound with
the S,S,S core configuration is the better inhibitor, but
the effect is marginal. Furthermore, despite indications
to the contrary provided by theoretical studies,35 it
appears that the presence of hydroxy, formyl or acetyl
substituents on the aromatic ring, which might act as H-
bonding donors or acceptors and considerably lower the
inhibitors’ logPo/w value, does not improve the IC50 (see
compounds TS-100–TS-105).
Inhibition of HIV-1 aspartic protease
Determination of IC50 values on recombinant wild-type
HIV-PR showed that the hexapeptide and pentapeptide
compounds are all quite efficient in inhibiting the pro-
tease (Table 1). Values measured for three commercial
inhibitors that present similar non-hydrolysable cores
are also shown for comparison. Under our enzyme
assay conditions, the hexapeptide shows a value com-
parable to that of the commercial inhibitors but is still
quite large (Mr=825). Comparison with the analogous
diaminodihydroxy inhibitor TS-7021 (Scheme 4) indi-
cates that the monohydroxy isostere provides margin-
ally better binding to HIV-PR. Replacing the Kyn-Val

A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
4723
Inhibition of C. albicans Sap2
Inhibition by our compounds of another aspartic pro-
tease, Sap2, was also determined, using a protease puri-
fied from the H12 strain of C. albicans.21 As the fungal
Sap are monomeric proteins with two equivalent
domains,36 they are similar to the host aspartic pro-
teases, so that they can be used as models to test the
specificity of peptidomimetic inhibitors for the retroviral
protease.
Scheme 4. Structure of the inhibitor TS-700, based on a dihydroxy-
ethylene isostere in the central module P1–P1 .19ꢀ21
In effect, there does seem to be a considerable difference
in the way the reported inhibitors interact with the fun-
gal protease. The peptidomimetic TS-91, which extends
Molecular modelling
0
from P3 to P3 and is a potent inhibitor of HIV-1 PR,
An explanation for the observed variations in the IC50
of selected peptidomimetics for Sap2 was obtained by
molecular modelling. Docking of the S,S,S epimer TS-
98 was carried out onto the structure of the inhibitor A-
70450, bound to Sap2X.36 This resulted in a good over-
lap with the central part of A-70450, which is in contact
also inhibits Sap2, albeit in the low micromolar range
(IC50, 12 mM). However, comparison of TS-91 with the
dihydroxy analogue TS-70 (IC50, 0.2 mM, Scheme 4),21
indicates that in this case the dihydroxy isostere binds
more strongly to the fungal protease than the mono-
hydroxy isostere. Furthermore, regarding the stereo-
chemistry, it is apparent that for this enzyme the S,S,S-
configuration is preferred (see compounds TS-94–TS-
105, Table 1). Finally, contrary to the viral protease,
dmPoa is decidedly not suitable in P2, as indicated by
the high IC50 values of compounds TS-92–TS-97.
Again, the methyl groups on the Poa ring seem to play a
key role, that is however detrimental to binding for the
fungal protease, so that Poa, as well as its derivatives
with H-bond donor/acceptor groups, such as aPoa or
fPoa, appear to be more suitable for positions P2 and
0
with the S1 and S1 enzyme subsites (Fig. 1A). After
optimisation of the complex between TS-98 and the
enzyme, a molecular dynamics run went to equilibra-
tion, indicating a good fit of TS-98 to the active site
(Fig. 1B). In analogy with the reference inhibitor A-
70450, the hydroxy group of TS-98 forms H-bonding
0
interactions with catalytic Asp32 and Asp218, the P1
amide nitrogen forms H-bonding interactions with
Asp218 and the Poa carboxy group in P2 interacts with
the main-chain nitrogen of Gly85 (Fig. 1B). This allows
the inhibitor Phe side chain in P1 to enter a hydrophobic
pocket lined by Ile30 and Ile119. The S,R,S epimer, TS-
99, was more difficult to dock onto the reference inhi-
bitor, and after optimization and molecular dynamics, it
was apparent that two of the three sets of H-bonding
0
P2 . Indeed, compound TS-98, with a Mr of only 553,
has an IC50 value (4 mM), which is lower than that of
the hexameric inhibitor TS-91, indinavir and nelfinavir,
and is comparable to that of ritonavir.
Table 1. Inhibition of HIV-1 and C. albicans aspartic protease by hydroxyethylamine-based peptidomimetics.a
0
0
0
b
P3-P2-P1-É-P1 -P2 -P3
Mr
logPo/w
IC50
(Da)
HIV-PR
Sap2 (mM)
TS-91
TS-92
TS-93
TS-94
TS-95
Kyn-Val-Phe-É(S,R,S)[OH]-Phe-Val-Kyn
Kyn-Dtg-Phe-É(S,R,S)[OH]-Phe-dmPoa
Kyn-Thr-Phe-É(S,R,S)[OH]-Phe-dmPoa
Kyn-Val-Phe-É(S,S,S)[OH]-Phe-dmPoa
Kyn-Val-Phe-É(S,R,S)[OH]-Phe-dmPoa
825
756
718
716
716
5.8
7.2
4.3
5.2
5.2
4 nM
3 nM
4 nM
3 nM
3 nM
12
80
120
>50
140
TS-96
TS-97
TS-98
TS-99
TS-100
TS-101
TS-102
TS-103
TS-104
TS-105
dmPoa-Phe-É(S,S,S)[OH]-Phe-dmPoa
dmPoa-Phe-É(S,R,S)[OH]-Phe-dmPoa
Poa-Phe-É(S,S,S)[OH]-Phe-Poa
Poa-Phe-É(S,R,S)[OH]-Phe-Poa
hPoa-Phe-É(S,S,S)[OH]-Phe-hPoa
hPoa-Phe-É(S,R,S)[OH]-Phe-hPoa
aPoa-Phe-É(S,S,S)[OH]-Phe-aPoa
aPoa-Phe-É(S,R,S)[OH]-Phe-aPoa
fPoa-Phe-É(S,S,S)[OH]-Phe-fPoa
fPoa-Phe-É(S,R,S)[OH]-Phe-fPoa
609
609
553
553
584
584
636
636
608
608
7.6
7.6
5.6
5.6
4.1
4.1
5.0
5.0
5.0
5.0
28 nM
25 nM
5 mM
17 mM
7 mM
50
>50
4
>50
25
>50
5
30
9 mM
19 mM
23 mM
68 mM
80 mM
5
12
Ritonavirc
Indinavirc
Nelfinavirc
721
614
568
4.0
4.0
4.1
3 nM
1.3 nM
1.9 nM
1.2
50
234
0
0
aThree letter codes are used to identify aminoacids in P1/P1 , P2/P2 . Other abbreviations: Kyn=kynurenic acid;Dtg= d-a-(2-thienyl)glycine;
Poa=phenoxyacetic acid;dmPoa=2,6-dimethylphenoxyacetic acid;hPoa=2-hydroxyphenoxyacetic acid;aPoa=4-acetlyphenoxyacetic acid;
fPoa=4-formylphenoxyacetic acid. See also Scheme 3.
bCalculated using the cLogP programme (v 4.01, www.daylight.com).
cRitonavir, indinavir and nelfinavir were extracted from drug formulations and purified by RP-HPLC.

4724
A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
Figure 1. Model structure of TS-98 bound to the active site of C. albicans Sap2. The enzyme structure was based on that of PDB entry 1zap. (A) TS-
98 was first docked onto the original inhibitor so as to give a maximum overlap. A-70450 is in grey and TS-98 in red. (B) The original inhibitor was
then removed from the active site and the model structure of Sap2 bound to TS-98 minimized and subjected to a molecular dynamics run. H-
bonding interactions of TS-98 with enzyme residues, and Ile residues lining a hydrophobic pocket for the Poa group are shown. (C) TS-98 was then
0
replaced by TS-96 and the process repeated;TS-98 is in red, TS-96 is in grey. The region where a methyl group from the P 2 dmPoa residue in TS-96
sterically clashes with the active site is indicated by an arrow.
interactions, namely that between the P0oa carbonyl and
Gly85 nitrogen and that between the P1 amide nitrogen
and Asp218 (Fig. 1B), were broken;at the same time
neither of the inhibitor Phe side chains could fit into the
S1 hydrophobic pocket. The loss of these interactions
lead to a calculated complexation energy for the S,R,S
epimer less favourable by 6.8 kcal/mol, consistent with a
lower affinity of this configuration for the binding site.
This appears to be a general feature as a lower affinity
of the S,R,S epimer was also observed for the other
tetrameric inhibitors (Table 1).
residues Thr221, Thr222, Ile223 and Tyr225 (Fig. 1C).
These interactions, together with the enhanced exposi-
tion to solvent of the aromatic ring in P2, and a severe
strain induced in the whole protein structure, causes a
loss of calculated complexation energy of more than 40
kcal/mol with respect to TS-98. The binding site of Sap2
0
is narrower than that of HIV-PR, and the S2,S2 subsites
impose more stringent requirements on residues that
can be accepted. In this respect, the S2 sub-site in the
retroviral protease appears to be quite tolerant, and
potent inhibitors result from the incorporation in P2 of
different residues such as dmPoa, Thr, Val or Dtg
(Table 1).
Comparison of model structures for TS-98 and TS-96
bound to the enzyme indicated that the methyl groups
on the aromatic ring in the latter severely interfere with
binding to the Sap2 active site. In correspondence with
P2, steric interactions are relieved by a rotation of the
dmPoa moiety and its extension outwards from the
Conclusions
We have extended a highly flexible method for rapidly
assembling diverse aspartic protease inhibitors, which
could be generally advantageous for rapidly generating
0
binding site, while P2 is constrained at the cost of
bumping of one methyl group in particular with enzyme

A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
4725
Table 2. Yields, MS and HPLC data
ification (up to 4 mg per elution) was carried out either
on a Waters Radial-PAK C18, 8ꢃ10 mm cartridge
using a gradient of 30–80% of CH3CN in 0.05% aqu-
eous TFA in 50 min, and a flow of 1.5 mL/min, after
extraction, (TS-91–TS-95), or by directly injecting the
reaction mixture into a Waters Sep-Pak Plus disposable
C18 cartridge, using a gradient of 20–70% of CH3CN in
0.05% aqueous TFA in 100 min, and a flow of 0.8 mL/
min, (TS-96–TS-105). A Gilson UV/VIS 155 detector
set at the wavelength of 214 and 254 nm was used in all
the analyses.
Compd
Yield (%)
ES-MS (m/z)
HPLC (min)
TS-91
TS-92
TS-93
TS-94
TS-95
TS-96
TS-97
TS-98
TS-99
TS-100
TS-101
TS-102
TS-103
TS-104
TS-105
10
10
10
15
15
34
25
54
45
38
30
21
20
43
37
825.7
757.4
718.7
717.4
717.4
610.4
610.3
553.2
553.2
585.4
585.3
637.2
637.2
608.9
609.3
30.5
39.0
29.6
34.0
36.5
41.8
42.8
34.6
34.3
25.2
28.4
27.7
27.6
28.1
27.6
Synthesis of diaminoalcohol isosteres
Diaminoalcohols 4 and 6 were synthesized as descri-
bed.22 Reagents and conditions are summarized in
Scheme 2.
Synthesis of peptidomimetics
lead compounds active against aspartic proteases of
pathogens, including retroviral and secreted fungal pro-
teases or the digestive proteases of the Plasmodium fal-
ciparum.37 It may also be useful in probing for solutions
to the problem of resistance in HIV-PR mutants. In this
respect, it has been shown that mutations that lead to
resistance often affect the hydrophobic enzyme residues
Peptidomimetics were assembled in solution from the
central modules 4 and 6, on a 0.1 mmol scale, using
standard peptide synthesis protocols as described pre-
viously.20 Amino acidic or carboxylic acid residues, or
pre-synthesised dipeptides were added to the central
module (2 equiv per free amine group on modules 4 or
6), in the presence of equimolar amounts of TBTU and
HOBT and a 4-fold excess of DIPEA, in DMF. Syn-
thetic intermediates, as well as the hexameric and pen-
tameric final products TS-91–TS-95, were extracted
from 5% NaHSO4 with ethyl acetate (EtOAc), washed
with brine, dried over MgSO4, evaporated under
vacuum and recrystallised from EtOAc/hexane prior to
semipreparative RP-HPLC purification, using Waters
Delta-Pak C18 cartridges. The tetrameric products TS-
96–TS-105 were instead purified without extraction, by
directly loading the reaction mixture onto disposable
Waters Sep-Pak cartridges adapted for use on HPLC
equipment. All products were confirmed to have the
correct structure and to be pure by analytical RP-
HPLC/mass spectrometry. Yields from modules 4 and
6, mass spectral data and HPLC details are reported in
Table 2.
0
interacting with the inhibitor residues in the P3/P3
positions7 and that molecules without a P3 group or
with one of reduced size exhibit a strong inhibition
against both wild-type and drug resistant mutant pro-
teases. Our tetrameric inhibitors based on the dmPoa
moiety thus appear to be excellent candidates for fur-
ther development.
Experimental
General
Fmoc-protected amino acids, benzotriazol-1-yl-oxy-
tris(pyrrolidino)-phosphonium
hexafluorophosphate
(PyBOP) and 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetra-
methyluronium tetrafluoroborate (TBTU) were from
ABI (Columbia, MD, USA). N,N-Diisopropylethyla-
mine (DIPEA), dimethylsulfoxide, dimethylformamide
(DMF) hydroxybenzotriazole (HOBT), phenoxyacetic
acid (Poa), 2-hydroxyphenoxyacetic acid (hPoa), 4-for-
mylphenoxyacetic acid (fPoa), 4-acetylphenoxyacetic
acid (aPoa) and kynurenic acid (Kyn) were from Fluka
Chemie (Buchs, CH), d-a-(2-thienyl)glycine (Dtg),
Fmoc-N-hydroxysuccinimide ester (Fmoc-OSu), N-
(tert-butoxycarbonyl)-l-phenylalanine methyl ester, and
all other reagents used in the synthesis of 4 and 6 (see
caption to Scheme 2) were from Sigma-Aldrich (St
Louis, MO, USA). All other reagents were technical or
analytical grade as appropriate. Syntheses were mon-
itored using ES–MS spectrometry (API-I Sciex) which
was also used to determine the correct structure and
purity of each final product in conjunction with analy-
tical RP-HPLC (Water Symmetry C18, 3.5 mm, 4.6ꢃ50
mm) carried out on a Gilson instrument, using a 20–
60% gradient of CH3CN in 0.05% aqueous TFA in 30
min, and a flow of 0.8 mL/min. Semi preparative pur-
Synthesis of symmetric products
TS-91 and TS-96–TS-105 were prepared in one-pot
reactions by adding 4 equiv each of the pre-synthesized
dipeptide Kyn-Val-OH and TBTU or the appropriate
Poa derivatives to the central modules 4 or 6, previously
deprotected from the Boc group by treatment with 95%
TFA.
Synthesis of asymmetric products
TS-94 and TS-95 were synthesised by addition of the
dipeptide Kyn-Val-OH to the unprimed side, followed
by Boc deprotection and addition of dmPoa to the
primed side. The dipeptide was prepared by reacting
Kyn with H-Val-OtBu and then removing the protect-
ing group with 95% TFA. When using the dipeptide for
couplings, to reduce racemization, PyBop and collidine
were used in preference to TBTU and DIPEA.

4726
A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
Synthesis of TS-92 and TS-93
always used after running 20 initial steps of Simplex
linear minimization. All the calculations were carried
out in a continuum dielectric of relative permittivity
e=4rij.44
These compounds were obtained by adding Fmoc-
Thr(tBu)-OH or Fmoc-Dtg-OH (2 equiv) to the
unprimed side, extracting and deprotecting from Boc
with 95% TFA, adding dmPoa (2 equiv) to the primed
side, extracting and deprotecting from Fmoc with 95%
piperidine in DMF and then adding Kyn (2 equiv) to the
unprimed side. Fmoc-Dtg-OH, used in compound TS-92,
was previously prepared by treating Dtg with 0.8 equiv
Fmoc-O-succinammide and extracting as described above.
The structures of the two S,S,S and S,R,S epimers TS-
98 and TS-99, and the structure of the dimethyl-Poa
derivative TS-96 were built with standard bond lengths
and angles, minimized and submitted to an extensive
conformational search driven by a genetic algorithm
operating on all the rotatable bonds.45 The energies of
the absolute conformational minima thus obtained were
taken as reference energies of the free ligands.
Enzyme assays
Recombinant HIV-1 protease was purchased from
Bioczech (Prague, Czech Republic) and Sap2 was pur-
ified from C. albicans H12, as reported.21 The fluoro-
genic substrates Abz-Thr-Ile-Nle-Phe(NO2)-Gln-Arg-
NH2 (HIV-protease) and Arg-Glu-[5-(2-aminoethyl-
amino)-1-naphthalenesulfonate]-Ile-His-Pro-Phe-His-
Leu-Val-Ile-His-Thr-Lys-[40-dimethylaminoazobenzene-
4-carboxylate]-Arg (C. albicans protease) were from
Bachem (Bubendorf, CH) and Molecular Probes Eur-
ope (Leiden, NL), respectively. The extent of inhibition
of the activity of the aspartic protease of HIV-1 and C.
albicans (Sap2) was evaluated as previously reported for
compounds based on a dihydroxyethylene module20 and
for some of the inhibitors described here.21 In partic-
ular, assays of HIV-PR activity were performed at 25 ꢁC
in 400 mM NaCl, 1 mM EDTA, 1 mM DTT, 100 mM
MES–NaOH buffer, pH 5.5, and 30 mM fluorogenic
The ligands were then manually docked onto the struc-
ture of the A-70450/Sap2X complex so as to achieve
maximum overlap at the level of the fundamental
hydroxy group and of the core hydrogen-bonded and
hydrophobic groups. After removal of the reference
inhibitor, the new model structures were first optimised
by running an energy minimisation on the full structure
to a 0.05 kcal/mol/A energy gradient. The models
were then subjected to a molecular dynamics run in the
NTV ensemble, at 300 K for 500 ps. The complexes
resulted stable at the dynamic simulation, and equili-
bration was always achieved within 300 ps. The lowest
potential energy equilibrium geometries were finally re-
optimised to a final gradient of 0.001 kcal/mol/A.
Acknowledgements
substrate
Abz-Thr-Ile-Nle-Phe(NO)2-Gln-Arg-NH2.
Titration of HIV-PR was carried out with two reference
inhibitors, ritonavir and indinavir, using the equation
IC50=[E]/2+Ki (1+[S]/KM),38 where Ki are the pub-
lished inhibition constants for the reference inhibi-
tors7,39,40 and KM was 25 mM, as determined under our
conditions. The active enzyme concentration in the
assay, [E], was estimated to be 3.6ꢄ2.3 nM.
We thank Drs. S. Miertus and V. Frecer for valuable
suggestions. We would like to acknowledge grants from
the ‘Fondo Trieste’ and the Istituto Superiore di Sanita/
National Research Program on AIDS, and FIRB con-
tract RBNE01TTJW.
References and Notes
Molecular modelling
1. Erikson, J.;Neidhart, D. J.;VanDrie, J.;Kempf, D. J.;
Wang, X. C.;Norbeck, D. W.;Plattner, J. J.;Rittenhouse,
J. W.;Turon, M.;Wideburg, N.;Kohlbrenner, W. E.;Sim-
Calculations were carried out on Silicon Graphics
Octane 1 workstations. Sybyl 6.8 (Tripos Inc., 1699
South Hanley Road, St. Louis, MO 63144, USA;
www.tripos.com) was used for all the molecular
mechanics and molecular dynamics simulations. The
initial coordinates for the Sap2 structure correspond to
those of the complex of A-70450 with C. albicans
Sap2X,36 and were obtained from the Protein Data
Bank, Brookhaven National Laboratory (PDB entry
1zap). After adding hydrogens and removing crystal-
lographic water molecules, the starting model of the A-
70450–Sap2X complex was obtained by the multistage
minimization procedure described by Levit and Lifson41
and previously used also by other authors on Sap2X.42
The Cornell’s version of the AMBER force-field,43 as
implemented in Sybyl 6.8, was used in all energy mini-
misations and dynamics runs. The starting structure, as
well as all other structures of ligands and enzyme com-
plexes, was minimized until a convergence criterion of
0.001 kcal/mol/A was achieved for all energy gradients.
The conjugate gradient minimisation algorithm was
mer, R.;Helfrich, R.;Paul, D. A.;Knigge, M.
249, 527.
Science 1990,
2. Fitzgerald, P. Curr. Opin. Struct. Biol. 1993, 3, 868.
3. Hoegl, L.;Korting, H. C.;Klebe, G. Pharmazie 1999, 54,
319.
4. Tomasselli, A. G.;Heinrikson, R. L. Biochim. Biophys.
Acta 2000, 1477, 189.
5. Leung, D.;Abbenante, G.;Fairlie, D. P. J. Med. Chem.
2000, 43, 305.
6. Richman, D. D. Nature 2001, 410, 995.
7. Brik, A.;Wong, C.-H. Org. Biomol. Chem. 2003, 1, 5.
8. Borg-von Zeppelin, M.;Meyer, I.;Thomssen, R.;Wu rzer,
R.;Sanglard, D.;Telenti, A.;Monod, M. J. Invest. Dermatol.
1999, 113, 747.
9. De Bernardis, F.;Arancia, S.;Morelli, L.;Hube, B.;San-
glard, D.;Scha fer, W.;Cassone, A. J. Infect. Dis. 1999, 179,
201.
10. Staib, P.;Kretschmar, M.;Nichterlein, T.;Hof, H.;
Morschhauser, J. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 6102.
11. Stewart, K.;Abad-Zapatero, C. Curr. Med. Chem. 2001,
8, 941.

A. Tossi et al. / Bioorg. Med. Chem. 11 (2003) 4719–4727
4727
12. Atzori, C.;Angeli, E.;Mainini, A.;Agostoni, F.;Micheli,
V.;Cargnel, A. J. Infect. Dis 2000, 181, 1629.
13. Kempf, D. J. Methods Enzymol. 1994, 241, 334.
Girolamo, A.;Agatensi, L.;Cassone, A. J. Clin. Microbiol.
1999, 37, 1376.
29. Smolensky, G.;Sullivan, P. A.;Cutfield, S. M.;Cutfield,
J. F. Microbiology 1997, 143, 349.
30. White, T. C.;Agabain, N. J. Bacteriol. 1995, 177, 5215.
31. Gruber, A.;Speth, C.;Lukasser-Vogl, E.;Zangerle, M.;
Borg-von Zepelin, M.;Dierich, R.;Wu rzner, R. Immuno-
pharmacology 1999, 41, 227.
32. Korting, H. C.;Schaller, M.;Eder, G.;Hamm, G.;Bo h-
mer, U.;Hube, B. Antimicrob. Agents Chemother. 1999, 43,
2038.
14. Tossi, A.;Antcheva, N.;Romeo, D.;Miertus, S.
Res. 1995, 8, 328.
15. Miertus, S.;Furlan, M.;Tossi, A.;Romeo, D.
Phys. 1996, 204, 173.
Peptide
Chem.
16. Benedetti, F.;Miertus, S.;Norbedo, S.;Tossi, A.;Zlado-
toidzky, P. J. Org. Chem. 1997, 62, 9348.
17. Frecer, V.;Miertus, S.;Tossi, A.;Romeo, D. Drug Design
Discov. 1998, 15, 211.
18. Tossi, A.;Antcheva, N.;Benedetti, F.;Norbedo, S.;
Miertus, S.;Romeo, D. Prot. Pep. Lett. 1999, 6, 145.
19. Benedetti, F.;Magnan, M.;Miertus, S.;Norbedo, S.;
Parat, D.;Tossi, A. Bioorg. Med. Chem. Lett. 1999, 20, 3027.
20. Tossi, A.;Bonin, I.;Antcheva, N.;Norbedo, S.;Benedetti,
33. Cassone, A.;De Bernardis, F.;Torosantucci, E.;Tacco-
nelli, M.;Tumbarello, M.;Cauda, R. J. Infect. Dis. 1999, 180,
448.
34. Alterman, M.;Bjo rsne, M.;Mu hlman, A.;Classon, B.;
Kvarnstrom, I.;Danielson, H.;Markgren, P.-O.;Nillroth, U.;
Unge, T.;Hallberg, A.;Samuelsson, B. J. Med. Chem. 1998,
41, 3782.
F.;Miertus, S.;Nair, A. C.;Maliar, T.;Dal Bello, F.;Palu
G.;Romeo, D. Eur. J. Biochem. 2000, 267, 1715.
,
21. Skrbec, D.;Romeo, D. Biochem. Biophys. Res. Commun.
2002, 297, 1350.
22. Benedetti, F.;Berti, F.;Norbedo, S. J. Org. Chem. 2002,
67, 8635.
23. Deeks, S. G.;Smith, M.;Holodniy, M.;Kahn, J. O.
JAMA 1997, 277, 145.
24. Patick, A. K.;Potts, K. E. Clin. Microbiol. Rev. 1998, 11,
614.
35. Frecer, V.;Miertus, S. Macromol. Chem. Phys. 2002, 203,
1650.
36. Abad-Zapatero, C.;Goldman, R.;Muchmore, S.;Hutch-
ins, C.;Stewart, K.;Navaza, J.;Payne, C.;Ray, T.
Sci. 1996, 5, 640.
Protein
37. Coombs, G. H.;Goldberg, D. E.;Klemba, M.;Berry, C.;
Kay, J.;Mottram, J. C. TRENDS Parasitol. 2001, 17, 532.
38. Cha, S. Biochem. Pharmacol. 1975, 24, 2177.
39. Martin, J. A.;Redshaw, S.;Thomas, G. J. Prog. Med.
Chem. 1995, 32, 239.
25. Bartlett, J. A.;DeMasi, R.;Quinn, J.;Moxham, C.;
Rousseau, F. AIDS 2001, 14, 1369.
26. Sham, H.;Kempf, D.;Molla, A.;Marsh, K.;Kumar, G.;
Chen, C.;Kati, W.;Stewart, K.;Lal, R.;Hsu, A.;Betebenner,
D.;Korneyeva, M.;Vasavanonda, S.;McDonald, E.;Saldi-
var, A.;Wideburg, N.;Chen, X.;Niu, P.;Park, C.;Jayanti,
V.;Grabowski, B.;Granneman, G.;Sun, E.;Japour, A.;
Norbeck, D. Antimicrob. Agents Chemother. 1998, 42, 3218.
27. Sham, H. L.;Betebenner, D. A.;Herrin, T.;Kumar, G.;
Saldivar, A.;Vasavanonda, S.;Molla, A.;Kempf, D. J.;
Plattner, J. J.;Norbeck, D. W. Bioorg. Med. Chem. Lett. 2001,
11, 1351.
40. Moyle, G.;Gazzard, B. Drugs 1996, 51, 701.
41. Levit, M.;Lifson, S. J. Mol. Biol. 1969, 46, 269.
42. Pranav Kumar, S. V.;Kulkarni, V. M. Bioorg. Med.
Chem. 2002, 10, 1153.
43. Cornell, W. D.;Cielpak, P.;Bayly, C. I.;Gould, I. R.;
Merz, K. M., Jr.;Ferguson, D. M.;Spellmeyer, D. C.;Fox,
T.;Caldwell, J. W.;Kollmann, P. A. J. Am. Chem. Soc. 1995,
117, 5179.
44. Orozoco, M.;Laughton, C. A.;Herzyk, P.;Neidle, S.
Biomol. Struct. Dyn. 1990, 8, 359.
45. Judson, R. S.;Rabitz, H. Phys. Rev. Lett. 1992, 68, 1500.
J.
28. De Bernardis, F.;Mondello, F.;Scaravelli, G.;Pachı , A.;